Interest in the taxonomy of the nondiphtheriae corynebacteria has increased, with a resultant reclassification of earlier defined taxa and discovery of new pathogens in the group (1,8). One of the more commonly reported human infections with these organisms is infectious endocarditis. Since 1966, several case reports and reviews have described 191 cases of nondiphtheriae corynebacterial endocarditis on both prosthetic and native valves. In only six (3.1%) of these cases was association with an indwelling intravascular device specifically documented: two patients in association with chronic hemodialysis (6), one case attributed to an infected permanent pacemaker wire (9), two in connection with ventriculo-peritoneal shunts (6), and one in association with a ventriculo-atrial shunt (10). Of reports pertaining to hospital-acquired endocarditis in particular, in four case series totaling 80 cases, Corynebacterium species were not identified as causative agents (7,11–13). In another series, Corynebacterium species accounted for 1 (3.3%) of 30 cases (14), this case occurring postcardiac catheterization. A further series of 14 cases of hospital-acquired endocarditis describes 3 (21.4%) cases due to coryneform species but gives no species identification or source of infection (15).

We have described three cases of nondiphtheriae corynebacterial nosocomial endocarditis associated with IID, all occurring in a single hospital, accounting for one third of 10 cases occurring in an 18-month period. These cases include the first reported case of endocarditis caused by C. amycolatum. Of the remaining seven cases, two were due to methicillin-resistant S. aureus and five to methicillin-sensitive S. aureus.

Since Lehmann and Neumann proposed in 1896 that bacteria morphologically resembling the diphtheria bacillus be incorporated with it into the genus Corynebacterium, the classification of the coryneform bacteria has been drastically altered (8). This change has, in turn, increased the difficulty of identifying these organisms, as methods that reliably differentiate related species, such as mycolic acid chromatography, gas liquid chromatography, and molecular amplification techniques, cannot easily be used in the routine laboratory setting. During 1987 to 1995, 11 new Corynebacterium species were described (1). Commercial identification systems need to be updated to include those species relevant in human disease. C. amycolatum has only recently been included in the updated API Coryne database 2.0 (4), which, as illustrated by our case, may still lead to misidentification if used alone. That there have been no previous reports of C. amycolatum causing endocarditis may be due to its misidentification as other nonlipophilic fermentative corynebacteria species such as C. xerosis and C. minutissimum (3,16), both of which are associated with human disease.

Published material provides useful schema for differentiating C. amycolatum, C. minutissimum, and C. striatum by using colonial morphology, carbohydrate assimilation tests, and sensitivity to amoxycillin and the vibriostatic compound O/129, in conjunction with the API Coryne and API 20NE systems (17). Antibiotic sensitivity patterns may support identification, with C. amycolatum and C. jeikeium generally resistant to multiple antibiotics (18). In contrast, C. striatum, C. minutissimum, and C. xerosis generally are sensitive to a wide range of antibiotics.

Funke et al. (1) have published guidelines for identifying the coryneform bacteria, including simple phenotypic characteristics but also recommending more complex chemotaxonomic investigations and molecular genetic analysis if phenotypic characteristics do not differentiate between species. These guidelines are intended to facilitate the establishment of true disease associations of these organisms. In our experience, morphologic features combined with antibiogram profiles and the Coryne API allowed identification of two out of three corynebacteria (Table 2). As the colonial morphology and the Coryne API profile did not match in the isolate from Case 1, the organism was sent to our reference laboratory, where identification as C. amycolatum was confirmed by gas liquid chromatography (Microbial Identification System, Microbial ID, Newark, DE).

Treatment regimens described include penicillin alone, beta-lactam antibiotics or erythromycin plus gentamicin, and vancomycin. Our three patients all received dual therapy with vancomycin and rifampicin. This combination has not previously been described but was effective in two of the three patients (including the case of prosthetic valve endocarditis), who were both free from infection at least 5 months after cessation of therapy. In a review of 19 patients with prosthetic valve endocarditis due to diphtheroids (19), 4 (57.1%) of 7 patients treated with antibiotics alone were reported as cured at least 1 year posttreatment. Of 12 patients treated both medically and surgically, 7 (58.3%) were reported as having a microbiologic cure (19).

This study highlights the importance of the nondiphtheriae corynebacteria in severe human disease. Specifically, clinicians and microbiologists must be aware of the potential risk factors for nondiphtheriae corynebacterial endocarditis in the hospital setting and the danger of overlooking positive long-line tip cultures as normal skin flora or contaminants. Stringent identification of clinical isolates will be required to define the role of the nondiphtheriae corynebacteria in human disease.